Last-click attribution assigns 100% of conversion credit to the final touchpoint before purchase - structurally, almost always a demand-capture channel like brand search, direct, or retargeting. Upper-funnel channels like paid social, CTV, YouTube, and display create demand earlier in the journey and rarely receive the last click. As a result, last-click systematically under-credits awareness activity - not by a rounding error, but in many cases by an order of magnitude. Fospha's 2024 research found TikTok drove 788% more conversions than last-click reported, and Meta drove 226% more.
Picture the finance meeting where a senior leader pulls up the attribution report and asks a simple question: why are we spending so much on TikTok when the returns are a fraction of what branded search delivers?
It's a reasonable question if you're looking at last-click data. The branded search numbers look excellent. TikTok looks like a rounding error. The logical conclusion is to cut the channel that isn't performing and double down on the one that is.
The problem is that both conclusions are wrong, and acting on them sets up a growth stall that arrives quietly, months later, after the audiences that branded search is harvesting have thinned out because nobody was building demand at the top of the funnel.
Last-click doesn't show you what's working. It shows you what's standing closest to the sale. Understanding the difference is one of the most commercially consequential things a performance marketer can do.
Why does last-click under-credit upper-funnel channels?
Last-click attribution works by assigning 100% of conversion credit to the final interaction a customer has before purchasing. Simple, auditable, compatible with cookie and pixel tracking -and genuinely useful for one thing: understanding which demand-capture channels are closing transactions.
The problem is that modern ecommerce journeys don't work that way. A customer discovers your brand in a TikTok scroll, sees a YouTube pre-roll while watching a review, and a week later types your brand name into Google and buys. Under last-click, Brand PPC gets the credit. TikTok and YouTube get nothing, because they were never designed to receive the last click.
This creates structural bias. Awareness channels are architecturally invisible to last-click because:
They operate on impressions, not clicks. CTV ads cannot be clicked at all. Display and video influence is often view-based, with conversion following much later on a different device. Last-click requires a click in its chain; if there isn't one, the channel doesn't exist.
The journey crosses devices and time. A customer who discovers a brand on mobile social and converts five days later on desktop via a branded search creates a journey that last-click can't trace. The final session gets credit; everything that built intent disappears.
Demand-capture channels absorb the credit. Brand search and direct traffic are the primary beneficiaries of this mechanism. Brand search shows spectacular last-click ROAS because it intercepts demand that other channels created. A user who finds you on TikTok, then visits your site directly three days later, hands the win to Direct - not to the channel that introduced them to your brand. Fospha's TikTok halo effect quantifies this: Direct cannibalizes 44% of TikTok's true sales in last-click reporting.
How large is the gap between last-click and true channel contribution?
Last-click doesn't slightly undercount upper-funnel channels - it barely sees them at all.
Fospha's Peak Playbook 2025, analyzing 83,000 campaigns across $2.8bn in spend, found that last-click models undervalue paid channels by more than 90% on average, with traditional tools missing up to 92% of total marketing impact. Fospha's 2024 channel-level analysis across its client base showed Meta drove 226% more conversions than last-click attributed - rising to 101x more for Awareness and Consideration activity specifically - and TikTok drove 788% more, with upper-funnel TikTok activity showing 702x the conversions that last-click reported. YouTube and Demand Gen were undervalued by more than 14x. Results vary by brand, category, and channel mix, but the direction of the bias is consistent.
The pattern holds, and deepens, in more recent data. State of Retail Commerce 2026 Report found that last-click captures only 3% of Reddit's true revenue contribution, 5% of Pinterest's total impact, and just 4.17% of TikTok Smart+'s ROAS. These aren't niche channels - Reddit spend grew 245% YoY and Pinterest 131% in 2025, driven by brands that could see past what last-click was reporting.
Privacy changes have compounded the blind spot further. Signal loss from platform restrictions means a growing share of social conversions, particularly view-through and cross-device, never appear in click-based reporting at all, pushing more revenue into the unattributed "direct" bucket where last-click can mis-assign it.
What happens to brands that budget from last-click data?
The practical consequence is a self-reinforcing cycle that takes months to become visible, by which point it's expensive to reverse.
Last-click shows awareness channels underperforming. Budget moves toward demand-capture - brand search, retargeting, direct response. Demand-capture channels continue to look strong because they're harvesting existing intent. Over time, without upper-funnel investment, new intent typically thins - meaning demand-capture channels can look efficient right up until growth stalls. Prospecting pools narrow. Customer acquisition costs climb. ROAS on retargeting starts to soften as the audience it's fishing in stops being refreshed.
Growth plateaus. The team runs more tests on the bottom-funnel channels, looking for efficiency gains. Meanwhile, the upper-funnel channels that were driving the business sit underfunded or turned off, and nobody can prove it because the data that would show their contribution is invisible to the tools in use.
Fospha's State of Retail Commerce 2026 found that the top quartile of brands for blended ROAS spend 26% of their Paid Social budget on brand-building activity, compared to just 11% for the average brand. They spend 2x more on awareness and consideration overall. That gap is, at least in part, a measurement problem - one that last-click attribution tends to reinforce.
How does full-funnel measurement change what you can see?
The fix isn't abandoning last-click - it still provides useful signal for bottom-funnel optimization on high-intent, short-journey paths. The fix is adding measurement that sees what last-click cannot: impressions, view-through influence, cross-device journeys, and marketplace activity that pixel-based tools were never built to reach.
An always-on, impression-led Media Mix Model quantifies the full-funnel impact of every channel - from the TikTok scroll that introduced a customer to the branded search that closed them seven days later. It runs continuously, updating daily, so the insight arrives at the pace at which decisions need to be made. And because it ingests performance across DTC, Amazon, and TikTok Shop in a single model, cross-channel halo effects become visible: you can see how Meta spend drives Amazon conversions, and how TikTok awareness creates Direct revenue that last-click incorrectly credits to itself.
The reallocation results when brands switch from last-click to this kind of full-funnel view are consistent:
- Travelpro shifted budget from saturated lower-funnel spend toward demand creation delivering +19% blended ROAS improvement YoY, 4.8x growth in upper-funnel revenue, and a 22% reduction in CPM. Under last-click, the vast majority of that upper-funnel contribution was invisible.
- U Beauty used Fospha's Halo measurement to reveal a 20% ROAS uplift when Amazon sales were included alongside DTC - giving the team the confidence to scale peak spend 34% YoY, growing new customer conversions 10% and blended revenue 9%.
- Adanola shifted ~50% of Paid Social spend into upper-funnel activity across Meta, TikTok and Google, delivering 2.5x US revenue growth in 12 months, 48% UK ROAS uplift, and 96% increase in Meta ROAS.
These outcomes aren't available to teams running on last-click alone, because the signal that would have justified the investment simply wasn't visible.
What should you do differently as a result?
Treat last-click as a bottom-funnel optimization tool, not a budget allocation tool. It's genuinely useful for understanding which demand-capture channels are closing high-intent, short-journey transactions. It is not a reliable guide to which channels are creating demand, building brand, or driving the downstream conversions that appear in other channels later.
Run the diagnostic that reveals the gap. Compare your last-click channel rankings against a full-funnel view. Look at which channels have high assisted conversions but low last-click conversions - those channels are being robbed of credit. Cross-check platform-reported revenue against your actual backend numbers to size the underreporting. The channels at the top of your last-click report and the channels driving your actual growth are often different lists.
Bring your finance conversation into full-funnel data. The most important thing last-click does is lose your finance team's trust in upper-funnel investment. When branded search shows 6x ROAS and TikTok Awareness shows 0.4x, the budget conversation is already over - even if the TikTok number is wrong by an order of magnitude. Full-funnel measurement that can explain, transparently and daily, how demand creation channels convert downstream is what allows marketing to stop losing that conversation.
The channels you've been systematically under-investing in haven't failed. The measurement failed to see them.
Frequently Asked Questions
What is last-click attribution?
Last-click attribution assigns 100% of conversion credit to the final channel a customer interacted with before purchasing. It's compatible with cookie and pixel tracking and is the legacy default in most web analytics platforms. It's useful for understanding demand-capture channels on short, high-intent journeys and structurally misleading for anything that happens earlier in the funnel.
Why does last-click over-credit brand search?
Brand search captures intent that other channels created. When a customer discovers a brand on social, researches it over several days, then types the brand name into Google and buys - branded search receives 100% of the credit, even though in many cases it has intercepted demand that was already building. Incrementality research consistently finds that branded search shows only 20–40% true incrementality on average - meaning a significant share of those conversions would have happened through organic results anyway.
Can GA4's data-driven model fix this problem?
Partly. GA4's data-driven attribution distributes fractional credit across touchpoints and handles some cross-device behaviour within Google's ecosystem. But it remains click- and cookie-based, so it cannot see impression-only influence from CTV, display, or social view-through. It is also Google-biased, it only sees what flows through Google tags, and requires minimum conversion volumes to function, reverting to last-click below those thresholds. Upper-funnel channels remain structurally undervalued even in GA4's improved model.
What is view-through attribution and does it solve the problem?
View-through attribution (VTA) credits an impression when a conversion follows within a defined window, even without a click. It recovers some of the impact that click-only models miss and is essential for non-clickable channels like CTV. But it requires careful window management - overly wide windows inflate credit by capturing conversions that would have happened regardless - and should be validated against incrementality testing rather than taken at face value.
What does "credit theft" mean in attribution?
Credit theft refers to the mechanism by which demand-capture channels (brand search, direct, retargeting) appear to perform strongly in last-click because they receive attribution for demand that awareness channels created. Fospha's halo effect shows that Direct cannibalizes 44% of TikTok's true sales in last-click reporting - customers who discovered the brand on TikTok and converted via direct visit days later are counted as Direct conversions, not TikTok conversions.
How does impression-led measurement work?
An impression-led Media Mix Model ingests spend, impressions, and outcome data across all channels - including those that operate on a view-through or no-click basis - and uses statistical modeling to estimate each channel's contribution to revenue. Unlike click-based attribution, it can estimate the contribution of CTV, display, and social awareness activity. Fospha's always-on MMM updates daily at the ad level, providing the strategic depth of traditional MMM at the operational speed of attribution. The methodology is described in detail in Fospha's model spotlight.
What is the Amazon halo effect and why does last-click miss it?
The Amazon halo effect describes the phenomenon where paid media activity on channels like Meta or TikTok drives downstream sales on Amazon - sales that are invisible to pixel-based measurement tools that only see your .com. When a customer sees a brand ad on Instagram and buys the product on Amazon two days later, pixel attribution records no conversion. A unified measurement model that ingests both DTC and Amazon sales alongside all media channels can quantify this cross-channel halo. Fospha's research, covered in detail on ClickZ, indicates that roughly 40% of Amazon revenue for some brands is influenced by non-Amazon media.
Related reading:
